Statistical Language and Speech Processing

Name: Statistical Language and Speech Processing | Third International Conference, SLSP 2015, Budapest, Hungary, November 24-26, 2015, Proceedings
Brand: Springer
Price: 53.49 EUR
Availability: OnlineOnly

Third International Conference, SLSP 2015, Budapest, Hungary, November 24-26, 2015, Proceedings

Adrian-Horia Dediu Carlos Martín-Vide Klára Vicsi(Herausgeber*in)

Springer (Verlag)

Erschienen am 16. November 2015

XVII, 302 Seiten

E-Book

PDF mit Wasserzeichen-DRM

Systemvoraussetzungen

978-3-319-25789-1 (ISBN)

53,49 €inkl. 7% MwSt.

Systemvoraussetzungen

für PDF mit Wasserzeichen-DRM

E-Book Einzellizenz

Als Download verfügbar

Beschreibung

Weitere Details

Weitere Ausgaben

Inhalt

Intro
Preface
Organization
Invited Talks (Abstracts)
Low-Rank Matrix Learning for CompositionalObjects, Strings and Trees
Towards Two-Way Interactionwith Reading Machines
Contents
Towards Two-Way Interaction with Reading Machines
1 Introduction
2 Challenges
2.1 Injecting Explanations
2.2 Extracting Explanations
3 Conclusion
References
The Prediction of Fatigue Using Speech as a Biosignal
Abstract
1 Introduction
2 Background
3 Corpus Collection and Labelling
4 Feature Extraction
5 Model Construction Procedure
6 Using Gaussianized Features
7 Shifting the Classification Threshold
8 Split Validation
9 Conclusion
Acknowledgements
References
Supertagging for a Statistical HPSG Parser for Spanish
Abstract
1 Introduction
2 Related Work
3 Lexical Frames
3.1 Corpus
3.2 Verb Classes
4 Experiments
4.1 Lemmas
4.2 Lemmas and POS
4.3 Only Verbs with Arguments
5 Conclusions
References
Residual-Based Excitation with Continuous F0 Modeling in HMM-Based Speech Synthesis
1 Introduction
2 Related Work
3 Methods
3.1 Data
3.2 Analysis
3.3 HMM Training
3.4 Synthesis
4 Evaluation
4.1 Methods of the Subjective Experiment
4.2 Results of the Subjective Experiment
5 Discussion and Conclusions
References
Discourse Particles in French: Prosodic Parameters Extraction and Analysis
1 Introduction
2 Discourse Particles
2.1 Main Features
2.2 Illustration: quoi and voilà
3 Methodology and Corpus
3.1 Corpus Constitution and Extraction
3.2 Speech Data Pre-processing
4 Analysis of Results
4.1 Pauses
4.2 Position in the Intonation Group
4.3 Pitch Level and F0 Slope
4.4 F0 Slopes Linking with Left and Right Contexts
4.5 Vowel Duration
4.6 Automatic DP Identification
5 Conclusion
References
Effects of Evolutionary Linguistics in Text Classification
1 Introduction
2 Related Work
3 Data Description and General Approach
4 Empirical Study
5 EL-Framework for Text Classification
6 Experiments and Results
6.1 EL-Framework Evaluation
7 Conclusions and Future Work
References
Evaluation of the Impact of Corpus Phonetic Alignment on the HMM-Based Speech Synthesis Quality
1 Introduction
2 Description of the Synthesis Platform LIPS3
2.1 HTS
2.2 The Language Processing Module
2.3 The Corpus
3 Description of the Experiment: Labeling Variation
3.1 Label Addition and Suppression
3.2 Boundary Shifts
3.3 TTS System Training
4 Analysis of the Subjective Evaluations
4.1 Conditions of the Test
4.2 Results
5 Discussion
6 Conclusion
References
Decoding Distributed Tree Structures
1 Introduction
2 Background: Encoding Structures with Distributed Trees
3 Going Back: Reconstructing Symbolic Trees from Vectors
3.1 Standard CYK Algorithm
3.2 CYK over Distributed Trees
3.3 Additional Rules
4 Experiment
4.1 Setup
4.2 Results
5 Conclusions and Future Work
References
Combining Continuous Word Representation and Prosodic Features for ASR Error Prediction
1 Introduction
2 Related Work
3 Continuous Word Representations
3.1 Description of Embeddings
3.2 Continuous Word Representation Combination
4 Error Prediction System
4.1 Set of Features
4.2 Architecture
5 Experiments
5.1 Experimental Data
5.2 Experimental Results
6 Conclusion
References
Semi-extractive Multi-document Summarization via Submodular Functions
1 Introduction
2 Background on Submodularity
2.1 Definition
3 Related Work on Automatic Document Summarization
4 Semi-extractive Document Summarization
4.1 Preprocessing
4.2 Problem Formulation
4.3 Solving the Problem
5 Experimental Results
5.1 Comparison with the State-Of-The-Art
5.2 Experiments
6 Conclusion
References
A Comparison of Human and Machine Estimation of Speaker Age
Abstract
1 Introduction
2 Speech Corpus
3 Human Prediction Performance
4 Machine Prediction Performance
4.1 Feature Analysis
4.2 Machine Learning
4.3 Raw Prediction Performance
4.4 Effect of Balancing the Training Set
5 Discussion
References
Acoustical Frame Rate and Pronunciation Variant Statistics
1 Introduction
2 Speech Data and Modeling
3 Speech-Text Alignment
3.1 Lexicon and Pronunciation Variants
3.2 Example of Phone Segmentation
4 Impact of Frame Rate on Statistics
4.1 Mute ``e''
4.2 Analysis of Final Clusters
5 Conclusion
References
The Influence of Boundary Depth on Phrase-Final Lengthening in Russian
1 Introduction
2 Material
3 Method
3.1 Obtaining Data from the Corpus
3.2 Factors Included in the Analysis
3.3 Duration Normalization
3.4 Statistical Analysis
4 Results
4.1 Boundary Depth
4.2 Utterance-Final Lengthening
5 Discussion
References
Automatic Detection of Voice Disorders
Abstract
1 Introduction
2 Method
2.1 Pathological and Healthy Speech Databases
2.2 Pre-processing Methods
2.3 Parameter Selection
3 Classification Experiments
3.1 Classification of Healthy and Pathological Voices Based on RBH Scale
3.2 Accuracy of Healthy and Pathological Classification on Different Sized Training Sets
3.3 Classification of Voice Disorder Based on Diseases
3.4 Classification of Voice Disorders into Multiple Classes
4 Conclusions
References
Semantic Features for Dialogue Act Recognition
1 Introduction
2 Related Work
3 Semantics for Dialogue Act Recognition
3.1 Latent Dirichlet Allocation
3.2 Semantic Spaces
3.3 Dialogue Act Recognition
4 Evaluation
4.1 Corpus
4.2 Tools and Model Configuration
4.3 Dialogue Act Recognition with Manual Speech Transcripts
4.4 Dialogue Act Recognition with Automatic Speech Recognition
5 Conclusions and Future Work
References
Conversational Telephone Speech Recognition for Lithuanian
1 Introduction
2 Lithuanian Phonemic Inventory
3 Experimental Setup
3.1 Data Set
3.2 Baseline Recognition Systems
3.3 Performance Metrics
4 Experimental Results
5 Summary
References
Long-Term Statistical Feature Extraction from Speech Signal and Its Application in Emotion Recognition
1 Introduction
2 Feature Extraction
3 Proposed Method
3.1 Workflow
3.2 Advantages
3.3 Comparison with UBM-GMM
3.4 Dimension of the Supervector and Curse of Dimensionality
4 Experimental Results
4.1 System Setup
4.2 Performance Evaluation
4.3 Results and Discussion
5 Conclusion
References
Rhythm-Based Syllabic Stress Learning Without Labelled Data
1 Introduction
2 Methods
2.1 Rhythm-Based Automatic Label Generation
2.2 Acoustic Features
2.3 Learning Algorithms
2.4 Materials
3 Experiments
3.1 Learning with Automatic Labels
3.2 Learning with Different Classifiers
4 Discussion and Conclusions
References
Unsupervised and User Feedback Based Lexicon Adaptation for Foreign Names and Acronyms
1 Introduction
2 Detecting Special Vocabulary Units
3 Pronunciation Adaptation
4 User Feedback Based Adaptation
5 Experiments
5.1 System
5.2 Data
6 Results
6.1 Unsupervised Lexicon Adaptation
6.2 User Feedback Based Adaptation
7 Conclusions and Discussion
References
Combining Lexical and Prosodic Features for Automatic Detection of Sentence Modality in French
1 Introduction
2 Experimental Setup
2.1 Textual Data for Training Language Models
2.2 Speech and Textual Data for Modality Detection
2.3 Configuration
3 Features for Question Detection
3.1 Linguistic Features
3.2 Prosodic Features
4 Experiments and Results
4.1 Prosodic Features
4.2 Linguistic Features
4.3 Combined Prosodic-Linguistic Features
4.4 Combined Outputs
5 Conclusions
References
Corpus Based Methods for Learning Models of Metaphor in Modern Greek
1 Introduction
2 Method
3 Implementation
3.1 Corpus Collection and Preprocessing
3.2 Classification
4 Empirical Evaluation
5 Conclusion
References
Probabilistic Speaker Pronunciation Adaptation for Spontaneous Speech Synthesis Using Linguistic Features
1 Introduction
2 The Buckeye Corpus
3 Method Overview
3.1 Overall Method
3.2 Conditional Random Fields
3.3 Experimental Setup
4 Feature and Window Size Selection
4.1 Linguistic Feature Selection
4.2 Window Size Selection
5 Backend Experiments and Discussion
6 Conclusion and Future Work
References
Weakly Supervised Discriminative Training of Linear Models for Natural Language Processing
1 Introduction
2 Classifier Risk Approximation
3 Risk Minimization Algorithm
3.1 Closed-Form Risk Estimate
3.2 Weakly Supervised Algorithm
4 Target Tasks
4.1 Task 1: Predicate Identification
4.2 Task 2: Entity Recognition
5 Results and Discussion
5.1 On the Gaussianity Assumption
5.2 Study of the Risk Estimator
5.3 Experiments with Gradient Descent
5.4 Closed-Form vs. Numerical Integration
6 Related Work
7 Conclusion
References
Merging of Native and Non-native Speech for Low-resource Accented ASR
1 Introduction
2 Background of Acoustic Modelling for Cross-Lingual or Accented ASR
2.1 Subspace Gaussian Mixture Models
2.2 Deep Neural Networks
3 Experimental Setup
3.1 Data
3.2 Baseline Systems
4 Language Weighting for Multi-accent Subspace Gaussian Mixture Models
4.1 Proposed Method
4.2 Results
5 Accent-Specific Top Layer for DNN
5.1 Proposed Method
5.2 Results
6 Conclusions
References
On Continuous Space Word Representations as Input of LSTM Language Model
1 Introduction
2 Model Architecture
2.1 Recurrent Neural Network Language Model
2.2 Skip-Gram and CBOW
2.3 Log-Bilinear Variant of Skip-Gram and CBOW
2.4 GloVe (Global Vectors for Word Representation)
2.5 LDA
2.6 Our Model Architecture
3 Experiments and Results
3.1 Data
3.2 Training
3.3 Results
4 Conclusion and Future Work
References
An Improved Hierarchical Word Sequence Language Model Using Word Association
1 Introduction
2 Review of HWS Language Model
3 Word Association Based HWS Model
3.1 Basic Idea
3.2 Training of NST
3.3 Converting Sentences by NST
3.4 Two Extra Techniques for HWS Model
3.5 Smoothing Methods for HWS
4 Intrinsic Evaluation
4.1 Settings
4.2 Results
5 Extrinsic Evaluation
5.1 Settings
5.2 Results
6 Conclusion
References
Neural-Network-Based Spectrum Processing for Speech Recognition and Speaker Verification
1 Introduction
2 Neural-Network-Based Acoustic Models
3 Neural-Network-Based Feature Extraction
4 Neural-Network-Based Feature Extraction for Speaker Verification
5 Experiments and Results
5.1 Speech Recognition
5.2 Speaker Verification
6 Conclusions and Future Work
References
Author Index

Systemvoraussetzungen

Als PDF speichern Als Link merken